Novel Mechanisms: What Has Been Tried
Part 5 of “The Tree of Liberty’s Roots: Repairing What Is Rotten”
Some reforms have strong evidence. Some have constitutional barriers. Some have been tried only in limited contexts, with results that are suggestive but not definitive.
This part examines novel mechanisms—democratic innovations that exist in limited experiments, advisory forms, or theoretical proposals. The evidence for these mechanisms is incomplete. The question is whether the incomplete evidence justifies broader experimentation.
Citizen Assemblies: The Strongest Evidence Among Novel Mechanisms
A citizen assembly—also called a citizens’ jury, deliberative mini-public, or citizens’ panel—operates on a simple principle: randomly select a representative sample of citizens, provide them with information and time, ask them to deliberate on a policy question, and produce recommendations for consideration.
How Citizen Assemblies Work
The process typically follows four steps:
1. Selection by sortition: Participants are selected randomly from the population, stratified by demographic characteristics to ensure representativeness.[^1]
2. Information and preparation: Participants receive briefings, expert testimony, and balanced materials on the policy question they are considering.
3. Deliberation: Participants discuss the issue, question experts, consider different perspectives, and work toward consensus or majority recommendations.
4. Recommendations and consideration: The assembly produces a report with recommendations for the legislature, which may or may not act on them.
What the Evidence Shows
Ireland’s Citizens’ Assembly (2016-2018): The assembly, composed of 99 randomly selected citizens and a chair, spent nearly a year deliberating on Ireland’s Eighth Amendment constitutional prohibition on abortion. In 2017, the assembly recommended amending the constitution to permit abortion access by a vote of 64% to 36%.[^2] In 2018, the Irish public voted 66% to 34% to repeal the amendment.
The relationship between the assembly recommendation and the referendum outcome is suggestive but not definitive. Did the assembly influence public opinion? Did it give politicians political cover to support a referendum? Did it change the framing of the debate? The evidence doesn’t allow firm conclusions.
Ireland’s Constitutional Convention (2012-2014): 100 citizens and 33 members of parliament deliberated on constitutional reforms, including same-sex marriage. The convention recommended putting same-sex marriage to a referendum, which passed 62% to 38% in 2015.[^3]
Again, the relationship between the assembly and the outcome is suggestive. Same-sex marriage was already gaining support in Ireland; the convention may have accelerated change, or it may have reflected change already underway.
British Columbia’s Citizens’ Assembly on electoral reform (2004): 160 randomly selected citizens studied electoral systems and recommended mixed-member proportional representation. The recommendation was put to a referendum—where it was defeated 58% to 42% in 2005, failing to meet the 60% threshold required for adoption.[^4]
This case illustrates a limitation: assemblies can recommend, but they cannot compel implementation.
The evidence base is limited. These are the most-cited cases, but they represent three data points. A comprehensive survey would require examining all citizen assemblies worldwide—the number convened, questions addressed, recommendations produced, and outcomes achieved. That denominator does not exist in the literature.
What We Don’t Know
The evidence for citizen assemblies raises as many questions as it answers:
– Was it the assembly or the timing? Ireland was changing on abortion and same-sex marriage before the assemblies. Would referenda have passed without assemblies?
– What questions work best? Assemblies have succeeded on moral issues (abortion, same-sex marriage). Do they work equally well on technical issues (budget policy, regulatory design)?
– What makes legislatures act? Assemblies are advisory. Why did the Irish government commit to acting on assembly recommendations while British Columbia’s government put recommendations to a referendum that failed?
– Under what conditions do they work? Ireland is small, relatively homogeneous, and had a political culture that supported deliberation. Do assemblies work in larger, more diverse, more polarized societies?
The Trade-offs
Advisory, not binding: Assemblies have no legislative authority. They can recommend but not decide. If legislatures break their commitment to bring recommendations to a vote, the assembly has no recourse.
Selection by sortition vs. election: Participants are chosen randomly, not elected. This ensures demographic representativeness but raises questions about democratic legitimacy—why should unelected citizens make policy recommendations?
Legislator ceding agenda-setting: Assemblies work when legislatures allow them to set the agenda on specific questions. If legislatures only refer boring or settled questions to assemblies, assemblies cannot address real problems.
Time and cost: A typical assembly takes 6-12 months and costs millions of dollars. The resource requirements are significant.
The Verdict
Citizen assemblies have the strongest evidence among novel mechanisms. In Ireland, they preceded successful referenda on abortion and same-sex marriage. But the evidence base is limited—few cases, confounding variables, incomplete understanding of mechanisms.
The pattern—not yet established as a general rule—is that assemblies work when: – Legislatures commit to acting on recommendations – The question is moral rather than technical – The political culture supports deliberation
Whether this pattern generalizes to other contexts is unknown.
Feasibility: High. Assemblies are advisory—no constitutional barrier. The question is whether legislatures will cede agenda-setting authority on meaningful questions.
Sortition: The Theoretical Case
Sortition—the random selection of legislators—is the opposite of elections. The ancient Athenian democracy selected most public officials by lottery. Modern democracies select none.
Sortition takes two forms:
1. Advisory: Sortition bodies that advise elected legislatures.
2. Legislative: Sortition bodies that make laws.
The Theoretical Case
Proponents argue that sortition:
– Eliminates campaign finance concerns: No elections, no fundraising. – Creates demographic representativeness: Random selection produces a legislature that matches the population’s composition. – Reduces polarization: Legislators don’t need to cultivate a partisan base.
Critics argue that sortition:
– Produces an inexpert legislature: Random selection matches the population’s distribution of knowledge—which for complex policy questions means significant expertise gaps. – Lacks democratic legitimacy: Participants weren’t elected; they have no mandate. – Has no accountability mechanism: If randomly selected legislators serve fixed terms and can’t be removed for incompetence, there’s no direct accountability.
The Evidence
The evidence for legislative sortition is theoretical. No modern democracy has randomly selected its entire legislature. Experiments with selected citizens as policy advisors—citizen assemblies—have succeeded, but those are advisory, not legislative.
Ancient Athens used sortition extensively, but the city-state excluded women, slaves, and non-citizens—approximately 80% of the population—and operated on a scale fundamentally different from modern nation-states. The comparison is tenuous.
The verdict: Sortition for legislative functions is elegant in theory and untested in practice. The theoretical concerns about expertise and legitimacy are significant. The evidence base does not exist.
Feasibility: Near zero for legislative sortition (would require constitutional amendment). Moderate for advisory sortition bodies (could be created by statute).
Recall Elections: The State-Level Evidence
Recall elections allow voters to remove elected officials mid-term. Nineteen states provide recall for state officials.[^5] The federal Constitution contains no recall provision.
How Recalls Work
Voters gather signatures—typically a percentage of votes cast in the previous election for that office. If the threshold is met, a recall election is held. Voters answer two questions: Should the official be removed? If so, who should replace them?
What the Evidence Shows
California, 2003: Governor Gray Davis was removed after signatures were gathered. The recall threshold in California is 12% of votes cast in the previous election for that office, not 12% of registered voters as is sometimes claimed—the relevant denominator for signatures is votes cast, which was approximately 7.7 million in 2002.[^6] Davis was removed by a 55-45 vote. Arnold Schwarzenegger won the replacement election with 48.6% of the vote.[^7]
This case illustrates a structural feature of California’s recall system: the replacement candidate can win with far less support than the recalled official received. Schwarzenegger’s 48.6% in the replacement race was less than Davis’s 44.6% support in the “remove” vote—but because the replacement was a plurality contest among many candidates, Schwarzenegger became governor with less support than Davis had when recalled.
Wisconsin, 2011-2012: Following the controversial Act 10 law limiting public worker collective bargaining, both Democrats and Republicans engaged in recall attempts. Thirteen state senators faced recall attempts—nine Democrats and four Republicans. Six Republicans were recalled; two Democrats were recalled.[^8]
This case illustrates that recalls can be used for policy disputes, not just misconduct or incompetence. The recalls were partisan responses to a controversial policy, not accountability mechanisms for official wrongdoing.
Colorado, 2021: State Senator Julie Gonzales narrowly avoided recall, with 50.4% voting to retain her seat.[^9] The recall was driven by policy disagreement over police reform and pandemic restrictions.
The Trade-offs
Democratic accountability: Voters can remove officials without waiting for the next election cycle.
Perpetual campaigning: Officials can face recall attempts from the beginning of their terms.
Policy rather than misconduct: Recalls can be used—and have been used—for policy disagreements, not just official wrongdoing.
Partisan weaponization: Well-funded interests can use recall as a political tool.
The Verdict
Recalls provide a mechanism for removing officials mid-term, but the mechanism is vulnerable to partisan use. The question is whether policy disputes should trigger removal, or whether recalls should be reserved for misconduct and incompetence. Current state systems generally allow recalls for any reason—including policy disagreement.
Feasibility: High for state/local recall (already exists in 19 states). Near zero for federal recall (would require constitutional amendment).
Direct Democracy: The Swiss Model
Direct democracy—citizens voting directly on policy questions—has three standard forms:
1. Referendum: Citizens vote on a law passed by the legislature. 2. Initiative: Citizens propose a law and vote directly on it. 3. Plebiscite: Citizens vote on executive or legislature proposals.
The Swiss Model
Switzerland conducts direct democracy systematically. Referendums are automatic for certain types of laws. Initiatives require 100,000 signatures collected within 18 months. Three to four federal votes occur annually, each with multiple referendums or initiatives.[^10]
The evidence shows mixed results:
Decentralization in practice: Swiss direct democracy allows cantons (states) to opt out of certain federal laws when voters in that canton reject them. The system creates coordination challenges.
Voter fatigue: Swiss voters face multiple votes annually. Participation rates average 45-50% for federal votes—lower than election turnout in most democracies.[^11]
Policy incoherence: Voters may simultaneously support tax cuts and increased services, voting down initiatives that raise taxes but not recognizing the fiscal consequences. Direct democracy does not guarantee coherent policy outcomes.
Minority rights concerns: Swiss direct democracy has produced restrictions on immigration and religious construction after majority votes.[^12] The majority can vote on the rights of minorities.
The U.S. Structural Problem
State-level direct democracy is common—27 states allow initiatives, 27 allow referendums, 19 allow recall.[^13] Federal-level direct democracy would require constitutional amendment. The Constitution vests legislative power in Congress, not in the people directly.
Feasibility: High for state/local direct democracy. Near zero for federal direct democracy.
Participatory Budgeting: The Municipal Evidence
Participatory budgeting—citizens directly deciding how to spend a portion of the public budget—is one of the most widely implemented democratic innovations globally. Porto Alegre, Brazil pioneered the model in 1989.
What the Evidence Shows
Porto Alegre: Participatory budgeting increased citizen participation and improved service delivery in low-income neighborhoods. The program expanded from 1% of the municipal budget in 1989 to 20% by 1999.[^14] Voter participation in participatory budgeting meetings increased from 2,500 in 1989 to over 35,000 by 2000, according to the World Bank.[^15] However, the World Bank source is a guide rather than an academic study, and Porto Alegre’s political context was unique.
New York City: In 2012, the City Council allocated funds for participatory budgeting in eight districts. By 2015, the program expanded to 24 districts. Participation was significant—approximately 50,000 citizens participated in 2015—but the program represented only a small fraction of the city’s total budget.[^16]
Paris: In 2014, Paris launched participatory budgeting, allocating funds for citizen-allocated spending. The program involved citizens and funded thousands of projects.[^17]
The Trade-offs
Genuine participation: Citizens have decision-making authority over actual public spending.
Limited scope: Participatory budgeting typically involves a small portion of the total budget.
Scale problem: The model works at the municipal level. Scaling to the state or federal level is untested.
The verdict: Participatory budgeting works at the municipal level with moderate evidence. The evidence does not yet support conclusions about state or federal implementation.
Feasibility: High for municipal implementation. Unclear for state/federal implementation.
What the Evidence Does and Does Not Show
The novel mechanisms examined in this part share a common feature: limited evidence.
Citizen assemblies have the strongest evidence—two Irish cases preceded successful referenda. But the evidence base is small, and the conditions that produced success may not generalize.
Recall elections have been used extensively at the state level. They provide accountability but are vulnerable to partisan use for policy disputes.
Direct democracy works in Switzerland but produces voter fatigue, policy incoherence, and minority rights concerns. Federal-level direct democracy is constitutionally barred.
Participatory budgeting works at the municipal level with moderate evidence. Scaling is untested.
Sortition is theoretically elegant but lacks any real-world evidence for legislative functions.
The common pattern: advisory mechanisms with moderate evidence (citizen assemblies, participatory budgeting), legislative mechanisms with minimal or no evidence (sortition, federal direct democracy), and mechanisms that work but with significant trade-offs (recall, state-level direct democracy).
Next: The Constitutional Question
Parts 1-5 have examined what works and what doesn’t. The reforms with the strongest evidence have moderate impact. The reforms with the highest impact face constitutional barriers.
The next part confronts the hardest question: what requires constitutional amendment, what can be done statutorily, and whether the constitutional structure itself can be changed.
Constitutional Reform: The Tree of Liberty’s Roots: Repairing What Is Rotten
Part 1: The Diagnosis: What’s Broken — Why the system can’t fix itself—and what the founders actually built
Part 2: What Works: Electoral Reform — RCV, open primaries, House expansion—what could pass tomorrow
Part 3: What Works: Removing Unfit Leaders — Votes of no confidence, recall, and why we have nothing comparable
Part 4: What Works: Anti-Capture Mechanisms — Campaign finance, lobbying reform, democracy vouchers—what the evidence shows
Part 5: Novel Mechanisms: What Has Been Tried — You are here
Part 6: The Constitutional Question — What requires amendment, what can be done statutorily, and whether change is possible
Part 7: A Proposed Framework — How to prioritize reforms—and whether the system can fix itself
← Part 4: What Works: Anti-Capture Mechanisms | Part 6: The Constitutional Question →
Notes
[^1]: Smith, Graham. Democratic Innovations: Designing Institutions for Citizen Participation (Cambridge University Press, 2009), 92-98.
[^2]: Irish Citizens’ Assembly, “Eighth Amendment of the Constitution: Final Report,” June 2018, https://citizensassembly.ie/en/How-Ireland-Voted/
[^3]: Constitutional Convention of Ireland, “Final Report,” March 2014, https://www.constitution.ie/en/report/
[^4]: British Columbia Citizens’ Assembly on Electoral Reform, “Making Every Vote Count: The Case for Electoral Reform in British Columbia,” December 2004, https://www.bcuc.ca/
[^5]: National Conference of State Legislatures, “Recall of State Officials,” https://www.ncsl.org/research/elections-and-campaigns/recall-of-state-officials.aspx
[^6]: California Elections Code § 11382. The threshold is 12% of votes cast for governor in the last election, not 12% of registered voters.
[^7]: California Secretary of State, “2003 Statewide Special Election Results,” https://www.sos.ca.gov/elections/prior-elections/statewide-election-results/2003-statewide-special-election
[^8]: Wisconsin Elections Commission, “Recalls,” https://elections.wi.gov/elections-voting/recall
[^9]: Colorado Secretary of State, “Recall Election Information,” https://www.sos.state.co.us/pubs/elections/Recalls/
[^10]: Swiss Federal Chancellery, “Political Rights,” https://www.bk.admin.ch/bk/en/home/dokumentation/polkitisches-system/politische-rechte.html
[^11]: Linder, Wolf, and Sean Mueller. Swiss Democracy: Possible Solutions to Conflict in Multicultural Societies (Palgrave Macmillan, 2021), 3rd ed., 135-157.
[^12]: Kriesi, Hanspeter. “Direct Democracy.” In The Oxford Handbook of Swiss Politics, ed. Patrick Emmenegger et al. (Oxford University Press, 2022).
[^13]: National Conference of State Legislatures, “Initiative, Referendum and Recall,” https://www.ncsl.org/research/elections-and-campaigns/initiative-referendum-and-recall-overview.aspx
[^14]: Wampler, Brian. “A Guide to Participatory Budgeting.” In Participatory Budgeting, ed. Brian Wampler (World Bank, 2012). Note: This is a World Bank guide, not a peer-reviewed study.
[^15]: Ibid.
[^16]: New York City Council, “Participatory Budgeting,” https://www1.nyc.gov/site/council/budget/participatory-budgeting.page
[^17]: Ville de Paris, “Le Budget Participatif de Paris,” https://www.paris.fr/budget-participatif